I actually spotted some useful info in Christian Bauer's VIC-II write-up(http://www.zimmers.net/cbmpics/cbm/c64/vic-ii.txt) regarding RAM locations 0 and 1 within a 6510 context.
Quote:
With a similar effect you can also write to RAM addresses 0 and 1 from the processor. They are normally not available as the internal data direction register and data register of the 6510 I/O port are mapped to these addresses, and the data bus drivers stay in tri-state on a write access. But the R/W line is set to low state (this can be explained as the I/O port has been integrated afterwards into the existing design of the 6502) and so the byte read by the VIC in the first clock phase is written to RAM. If you want to write a certain value to addresses 0 or 1 you only have to write an arbitrary value to these addresses and take care that the VIC read the desired value from RAM in the clock phase before.
So you would only see the address on the address bus for writing to memory location 0/1 and not the data to be written to this location.
It is a rather interesting anomaly where RAM locations 0/1 would be populated with the data that the VIC read in the previous cycle if you wrote to locations 0/1.
I wrote a assembly program to test this on the Vice emulator.
I started off by populating the first screen line with A's and then wait till the raster line approaches these A's. With the raster line at this location I do a couple of writes to memory location 1. According to the theory memory location 0 should be populated with an image line of an 'A'.
I then switch the location of screen memory to location 0. This will enable us to see memory locations 0 and 1 as the first 2 characters displayed on the screen.
Here is the assembly:
Code:
0000 SEI 78
0001 LDY #$07 A0 07
0003 LDA #$01 A9 01
0005 LDX #$27 A2 27
0007 LOOP STA $0400,X 9D 00 04
000A DEX CA
000B BPL LOOP 10 FA ; Populated the first screen line with 'A's
000D LOOP2 LDA $D011 AD 11 D0
0010 BMI LOOP2 30 FB
0012 LDA $D012 AD 12 D0
0015 CMP #$34 C9 34
0017 BNE LOOP2 D0 F4 ; Wait in a loop till we reached ratser line 52
0019 LDX #$03 A2 03
001B LOOP3 STY $0001 8C 01 00
001E DEX CA ; Write a value number of times to loc 1
001F BNE LOOP3 D0 FA ; Hopefully at last read we are in visible char region
; where VIC-II read a value at first phase
0021 LDA #$05 A9 05
0023 STA $D018 8D 18 D0 ; Move screen memory to location 0
; first two chars on screen is locations 0
; and 1
0026 LOOP4 BNE LOOP4 D0 FE
0001 LDY #$07 A0 07
0003 LDA #$01 A9 01
0005 LDX #$27 A2 27
0007 LOOP STA $0400,X 9D 00 04
000A DEX CA
000B BPL LOOP 10 FA ; Populated the first screen line with 'A's
000D LOOP2 LDA $D011 AD 11 D0
0010 BMI LOOP2 30 FB
0012 LDA $D012 AD 12 D0
0015 CMP #$34 C9 34
0017 BNE LOOP2 D0 F4 ; Wait in a loop till we reached ratser line 52
0019 LDX #$03 A2 03
001B LOOP3 STY $0001 8C 01 00
001E DEX CA ; Write a value number of times to loc 1
001F BNE LOOP3 D0 FA ; Hopefully at last read we are in visible char region
; where VIC-II read a value at first phase
0021 LDA #$05 A9 05
0023 STA $D018 8D 18 D0 ; Move screen memory to location 0
; first two chars on screen is locations 0
; and 1
0026 LOOP4 BNE LOOP4 D0 FE
At the bottom I have attached a screen shot how the screen looks like switching the location of screen memory to 0.
You will see that the second character from the left at the top row is a less than symbol which have the screen code $3C. $3C is the second image line of the character A, which is what we expect since we did the writes at scan line 52.
It would be interesting to know if you get the same result when running on a real C64.
Attachment:
test.png [ 153.6 KiB | Viewed 4201 times ]
). While arranging to inhibit read data they apparently found it easiest to inhibit write data as well.